[CP2K-user] Initalization costs too much time in CeO2 AIMD calculation

Hedley Dong dmh19... at gmail.com
Mon Sep 7 12:10:44 UTC 2020

  Thanks for your reply,  Marcella. 
I have found a solution. Once I change coordination input file into .xyz 
rather than .pdb, the initalzation time decreases a lot. I didn't know why, 
but it works.
Output file is attached, it may provide more information.

在2020年9月5日星期六 UTC+8 下午9:37:18<Marcella Iannuzzi> 写道:

> The direct comparing  of systems with different electronic properties is 
> not straightforward.
> There is no strong reason to expect the same behaviour. 
> CeO2 is known to be a difficult case,  see the  the literature. You might 
> need to adapt the model.
> Without output and timings it is anyway hard to say something and give 
> advise. 
> Do you restart the MD reading the wavefunction from the restart file as 
> initial guess?
> Kind regards
> Marcella
> On Saturday, September 5, 2020 at 5:02:20 AM UTC+2 dm... at gmail.com 
> wrote:
>> Hi,
>> I started an AIMD calculation of CeO2(110) slab, including 144 atoms and 
>> 2016 electrons. With current configuration, it costs more than 300s for one 
>> MD step. Compared to another system, Pd20 cluster on TiO2(110), including 
>> 451 atoms and 3810 electrons, with the same DFT configuration, it costs 
>> about 100s for one step. I'm using cp2k 7.1, OT method and ASPC as 
>> extrapolation method, and i found that it only cost less than 30s for SCF 
>> converged, yet 200s needed to start next MD step.
>> Is there any problems with CeO2 systems or my input files?
>> Thanks!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200907/2ad48fbd/attachment.htm>

More information about the CP2K-user mailing list